CV

My name is Aaditya Barak. I am a student of Electrical and Electronics Engineering at University Institute of Engineering and Technology, Chandigarh. I am a machine learning enthusiast and currently in third year of my undergraduate studies.

• Phone: +91-9896127290
• Email: aadityabarak472@gmail.com
• GitHub: https://github.com/AddyB0t
• X: https://x.com/aadi29494
• Hugging Face: https://huggingface.co/AddyB0t

Technical Skills

• Programming: Python, C, C++, MATLAB
• Machine Learning & AI: PyTorch, Hugging Face, OpenCV, CNN, NLP, FastAPI, LangChain, Model Fine-Tuning, StreamLit, RESTful API’s, WebSocket, Cursor, Claude Code, N8N
• IoT: Arduino, Raspberry Pi, ESP32, 8085 Microprocessor
• Tools: CX Programmer, NB Designer, P-Spice, Simulink, Cursor, Git/GitHub, Bash

Internship/Freelancing Experience

PRUDENTBIT (AI Intern)

Duration: May 2025 - July 2025

• Developed automated redaction for Excel files, enabling detection and replacement of sensitive information with entity placeholders to ensure data privacy and regulatory compliance
• Implemented a black box redaction system for documents, visually masking detected PII with precise, pixel-level overlays while preserving document structure and readability
• Built an advance image preprocessing pipeline that enhances scanned and low-quality documents, including skew correction, orientation adjustment and noise reduction, significantly improving OCR accuracy
• Designed an intelligent OCR workflow that segments images and maps extracted text to exact coordinates enabling accurate identification and redaction of sensitive data in both images and embedded PDF’s

ZapUp (Website)

• Developed a full-stack educational platform using Next.js, React and Supabase enabling dynamic access to hierarchical textbook content, exercises and AI-generated solutions for multiple subjects and classes
• Implemented an AI-powered study assistant chatbot with context-aware hints, educational content filtering and conversation history leveraging OpenRouter API for real-time student support
• Integrated Razorpay payment gateway for secure subscription management, including order creation, payment verification and automated user access control across multiple subscription tier
• Used Supabase to store the user preferences, state/school selection and real-time profile management with robust RESTful API endpoints

Projects

ML/AI Projects

Deepfake Detection System (Smart India Hackathon)

• Developed a CNN-based model achieving 93% accuracy using TensorFlow and VGG16 transfer learning
• Implemented face extraction preprocessing with OpenCV and comprehensive performance evaluation
• Created comprehensive Evaluation Metrics to measure model performance across diverse datasets
• Designed a solution to address growing concerns about AI-generated media manipulation

Token Classification With BERT-Large-Uncased

• Developed a Token Classification pipeline using a fine-tuned BERT-Large model, achieving high accuracy on the CoNLL-2003 NER dataset
• Leveraged Hugging Face Transformers and PyTorch for state-of-the-art NLP model implementation and training
• Integrated additional NLP preprocessing with SpaCy and FastAI for efficient data handling and training
• Evaluated model performance using standard NER metrics across diverse datasets with achieving accuracy 98.5%

Real-time Chat Room Application

• Developed a real-time chat application using FastAPI and WebSocket technology, implementing features like message persistence, user notifications, emoji support, and real-time communication protocols
• Created a responsive frontend interface with HTML/CSS/JavaScript, integrating real-time updates and user interactions, highlighting full-stack development capabilities and user experience design skills
• Used Ngrok for exposing the endpoint of FastAPI

Youtube Bot - AI-Powered

• Developed an AI-driven YouTube content assistant that integrates multiple language models (OpenAI, Anthropic, Google Gemini, and Hugging Face) to help with content creation and analysis
• Designed to assist with YouTube content management and generation, leveraging various AI services through the Langchain Framework for enhanced functionality

RAG ChatBot - Document-Based Chatbot System

• Built a sophisticated document-based chatbot using FastAPI backend and StreamLit frontend that processes multiple file formats (PDF, TXT, CSV, MD) and enables natural language queries against uploaded documents
• Implemented advanced embedding techniques with FAISS vector store for semantic search and integrated GPT-4 for intelligent, context-aware responses based on documents content
• Developed comprehensive system with PostgreSQL database SQLAlchemy ORM, exception logging, chat history tracking, and PII protection features
• Created RESTful API endpoint for document management, real-time chat interface, and monitoring capabilities with robust error handling and input validation

Research Paper Summarization Tool

• Created Streamlit web app using OpenAI’s language models through Lang Chain to summarize research papers
• Built user-friendly interface with customizable explanation styles for academic research papers
• Implemented robust error handling and preprocessing for various document formats
• Significantly reduced research review time while maintaining comprehension of key concepts

IOT Projects

Weather Station

• Designed a weather station using Arduino UNO, DHT11 sensor, and LCD display to measure temperature and humidity with 95% accuracy
• Implemented I2C communication protocol for efficient data transfer between components with real-time display capabilities

Alcohol Detection System

• Designed an alcohol detection device using Arduino UNO and MQ-3 sensor that triggers a buzzer alert when levels exceed a calibrated threshold
• Implemented serial monitoring for real-time tracking with 95% detection accuracy and appropriate sensitivity adjustments

Smoke Detection System

• Designed a smoke detection device using Arduino UNO and MQ-3 sensor with 95% detection accuracy through proper calibration and testing
• Implemented real-time monitoring with buzzer alerts when smoke levels exceed predetermined safety thresholds

Smart Dustbin

• Designed an automated waste container that opens when a person approaches using ultrasonic sensor for proximity detection up to 20cm
• Programmed Arduino to control servo motor with 2-second delay timing mechanism for proper lid operation

Blind Assistance System

• Designed a navigation aid for visually impaired individuals using Arduino UNO and HC-SR04 ultrasonic sensor to detect obstacles within 100cm
• Implemented variable feedback where buzzer beeps faster as obstacles get closer, with response time under 100ms for immediate feedback

Distance Measurement System

• Designed a precise distance measurement system using Arduino UNO and HC-SR04 ultrasonic sensor with I2C LCD display for real-time readings
• Achieved consistent performance with measurement range of 2-400cm, accuracy within ±1cm, and refresh rate of 2Hz

Password-Based Door Lock System

• Designed a secure door locking system using Arduino UNO, 4x4 matrix keypad, and servo motor with LED indicators for lock status
• Implemented error handling for incorrect attempts with 99% reliability and response time under 100ms for real-time security application

Certifications

• Lang Chain and SQL Intermediate Certifications (Udemy, Solo learn)
• Python Programming (EI Systems)
• SQL Intermediate Level (Sololearn)

Achievements

• Qualified for Smart India Hackathon 2024 in the college-level selection round

Positions of Responsibility

• Community Manager for an international Discord server of Machine Learning enthusiasts
• Lead Administrator for University’s E-Sports Discord community

Education